An Improved Flow Calorimeter. Experimental Vapor Heat Capacities

An Improved Flow Calorimeter. Experimental Vapor Heat Capacities and Heats of Vaporization of n-Heptane and 2,2,3-Trimethylbutane1. Guy Waddington ...
2 downloads 0 Views 1MB Size
22

GUYWADDINGTON, SAMUEL S. TODDAND HUGHM. HUFFMAN

concentration of I : 6, and thereafter a slight decrease. Similar behavior has been observed in solutions of zinc chloridezband hydrogen ch10ride.I~ If the relative intensity of the two components of the band can be taken as a measure of the extent of inter-molecular association in water, then water in a 3.7 d l solution of potassium thiocyanate a t 25' has a structural temperature around G O O , and in a 9.3 -11solution over looo. This striking effect of potassium ion and thiocyanate ion on the structure of water is in conformity with the low hydration of both ions and the resulting far extending ion-fields, as discussed by Ulich.14 (13) L. Ochs, J. Gueron and AI. Magat, J. pkys. r n d i u m , (8) 1, 85 (1940). (14) H. Ulich, Anpew. Clirm., 49, 270 (1936).

[COSTRIBUTION FROM

BUREAUOF

Vol. 69

Acknowledgment.-The author wishes to express his thanks to the China Institute in America, New York, for a scholarship covering the period during which the work here reported was carried out.

Summary The Raman spectra of aqueous solutions of potassium thiocyanate and of pure water a t various concentrations and temperatures have been described. The results show a great increase in the structural temperature of water in potassium thiocyanate solutions caused by the perturbing ionic field created by the solute. MADISOS,WISCOSSIN

RRCEIVED J U N E 20, 19-@

MINES, PETROLEUM EXPERIMENT STATION]

An Improved Flow Calorimeter. Experimental Vapor Heat Capacities and Heats of Vaporization of n-Heptane and 2,2,3-Trimethylbutane1 BY

GUY

WADDINGTON,'

SAMUEL

s. TODD3 AND HUGHM. HUFFMAN4

Introduction The dearth of reliable heat capacity values for vapors has been emphasized in recent years by the increasing need of the petroleum and allied industries for such information. Theoretical and quasi-theoretical calculations of heat capacities and other thermodynamic quantities from spectral data, by the methods of statistical mechanics, have done much to supplement and to extend existing experimental data. However, as the accuracy and self-consistency of thermodynamic values ultimately rest upon the quality of the available experimental evidence, it is evident that additional accurate measurements are desirable. To fill this need, the Petroleum and Natural Gas Division of the Bureau of Mines, U.S. Department of the Interior, has instituted, as part of its thermodynamic research program, a project for the development and application of accurate methods for determining heat capacities of hydrocarbon vapors and their derivatives. This paper describes the apparatus and experimental procedure now in use and also presents molal heat capacities and heats of vaporization for nheptane and 2,2,3-trimethylbutane. Modified versions of Swann'sj continuous flow method have given good results in the hands of careful experimenters. Of the early work, that of

Scheel and Heuse6 and of Osborne,' et al., may be mentioned, while recent contributors in this field include De Vries,8 Felsing,g P i t z e P and Scott I 1 and their co-workers. The method used by Osborne' and by Scottll reduced losses to a minimum by creating nearly adiabatic conditions in the calorimeter, while the others involved heat-loss corrections obtained from measurements a t several flow rates. In this Laboratory the design due to Pitzer'O was used as a starting point for the following reasons: (1) the heat losses may be made small and can be corrected for by a linear relationship if all important variables are taken into account; ( 2 ) it is well-adapted to the study of compounds the boiling points of which are above room temperature; (3) it permits the incorporation of changes of design and operating procedure, which may lead to improvements in precision. Very briefly, the method consists of passing a measured constant flow of vapor, generated by electrical evaporation in a cycling vaporizer, through a calorimeter in which the vapor is heated electrically by a measured and constant power input. The apparent heat capacity Cpoba. is given by E / F A T , where 2, T i s the temperature increase produced when E units of energy per unit .of time are imparted t o a vapor flowing at a rate of F moles per unit of time. The experi-

(1) Published by permission of the Director of the Bureau of Mines, U. S. Dept. of the Interior. Not copyrighted. (2) Physical Chemist, Bureaii of Mines, Petroleum Experiment Station, Bartlesville, Okla. (3) Associate Physical Chemist, Bureau of Mines, Petroleum Experiment Station, Bartlesville, Okla. (4) Principal Physical Chemist, Bureaii of Mines, Petroleum Bxperiment Station, Bartlesville, Okla. (5) Swann, Proc. R o y . SOC.(Ixndon), 82A, 127 (1909).

(6) K. Scheel and W. Heuse, A n n . Physik, [ 4 ] 37, 79 (1!112). ( 7 ) N. S. Osborne, H. F.Stimson and T. S. Sligh, N u l l . Birr. Slandnrds Sci. TechrzoL Pnpers, 20, 119 (1925). (8) James B . Montgomery and Thomas De Vries, THIS JOURNAL, 64, 2372 (1942). (9) Benjamin P. Dailey with W. A . Felsing, i b i d . , 66, 42 (1923). ( I O ) Kenneth S. Pitzer, i b i J . , 65, 2413 (1941). ( 1 1 ) Russell B. Scott and Jane W. Mellors, J . Rcrcarrli N u l l . Birr. Slaridavds, 34, 228 (1945).

Jail., 1947

HEATCAPACITIES OF 12-HEPTANE

I

AND 2,2,3-TRIMETHYLBUTANE

23

VACUUM

TC 1&2

W

l

Fig. 1.---Essential features of the apparatus for measuring heats of vaporization and the heat capacities of vapors.

mental data are precisely represented by the rek / F where Cp is the true lationship Cfiabs. = heat capacity. Hence a plot of v s . l / F extrapolated t o zero value of 1 / F gives the correct value of Cp. The fact that the experimental points of the present work deviate on the average from a straight line by less than * 0.057, is convincing evidence for the validity of the correction for heat losses. The design of the apparatus used in the present work is such that the precision of the measured heat capacities is within * O.lCY,.

c9+

Experimental Apparatus.--The construction and operation of the calorimetric system may be followed by reference to Fig. 1. For simplicity certain parts are omitted. The main features are: (1) the cycling vaporizer A, contained in a constant-temperature bath; (2) the calorimeter C situated in a second thermostated bath ; (3) the flow-calibration system E. Hydrocarbon vapors are produced a t a steady rate by boiling 150 to 250 rnl. of the liquid hydrocarbon in the vacuum-jacketed vaporizer A with

electrical energy from storage batteries. The heater is an 11-ohm helix of no. 24 AWG manganin wire with double glass-fiber insulation. The wire is close-wound on a glass spiral made from 4mm. Pyrex rod fused to the wall of the boiling vessel, which is cylindrical and has a length of 20 cm. and a diameter of 6 cm. Current and potential leads of no. 20 and no. 24 AWG silver wire, respectively, are taken from the heater a t points below the surface of the liquid hydrocarbon and leave the vaporizer a t the top, through tungsten seals, under the surface of the bath liquid from which they are isolated by small glass tubing. The hydrocarbon vapor, prevented from condensing by electrically heated flow lines, after flowing through the calorimeter C, is condensed, brought up t o the bath temperature by flowing through a. long helix and returned to the vaporizer. Heat exchange between the vaporizer and its environment is negligible when the bath, which is controlled to better than * 0.003O, is maintained a t a temperature about 0.2' higher than the temperature of the boiling hydrocarbon and when a pressure of less than mm. is maintained in the silvered jacket space. The

24

'

GUYWADDINGTON, SAMUEL S. TODD AND HUGHM. HUFFMAN

temperature diflerence between the bath and the boiling liquid or vapor is measured by the copperconstantan difference couple, TC1, which can be placed a t any depth in the deep thermocouplewell shown. Another copper-constantan couple, TCz, indicates the boiling temperature of the liquid, and a platinum resistance thermometer is used to measure the bath temperature. Adjusting the relative flow of water in the two condensers, so that a small fraction of the total condensation occurs in condenser D, returns the liquid t o the vaporizer system a t near the boiling temperature. Helium is maintained in the rest of the system. Pressure is controlled to better than * 0.1 mm. (as observed on a mercury manometer and also estimated from small boilingtemperature variations) by having a slow stream of helium continuously entering the system and escaping through a small capillary tube, alternately opened and closed by a soft rubber seat on he movable arm of a relay, actuated by a sulfuric acid nianostatI2 mounted in an ice-bath, thus minimizing pressure drifts due to temperature effects. For operation below atmospheric pressure the relay is in a continuously pumped vacuum desiccator. The flow calorimeter is situated in a second thermostated bath, which is controlled to * 0.002O at low temperatures and * 0.005O a t high temperatures. The flowing vapor is brought to within =t3.0' (indicated on copper-constantan difference couple, TC,) of the bath temperature before it enters the bath, by controlling the heat input to the flow line. I t then passes through a helix made from I! m. of 8-mm. i d . glass tubing and past the platinum resistance thermometer T1, which, by comparison with another platinum resistance thermometer (not shown) in the bath itself, gives necessary knowledge of the temperature equilibrium between the rapidly flowing vapor and the bath. The thin-walled Pyrex U-tube, silvered on the outside, containing the calorimeter heater H and two platinum resistance thermometers Tz and TB,is surrounded by a silvered jacket connected to the high-vacuum system. The heater consists of 4 m. of no. 30 AWG Nichrorne wire wound in three concentric helices spaced from one another by small glass rods. The two inner helices, in series with one.another, are in parallel with the outermost helix giving an effective resistance of about 24 ohms. The unit is spaced from the glass tube, which is 10 mm. i.d. a t this point, by a snugly fitting, steep-pitched helix of glass insulated no. 20 AWG copper wire, which also serves as an electrical lead. Bringing the current and potential leads out of the calorimeter aguimt the entering vapor stream tends to eliminate conduction losses along the leads. The potential antl current leads leave the glass flow (12) E. B. Hirshberg and E. H. Huntress, I n d . Eng C h r w , Ann1 E d . , 5, 344 (1933).

Vol. e9

line outside of the calorimeter proper through tungsten seals and are isolated from the bath liquid by glass tubing. Electrical energy is supplied to the heater by storage batteries. Baffles B1 and BBminimize radiation losses along the length of the tube, while Ba helps to mix the gas after heating. Thermometers TZand T3 are isolated from radiation from the heater by being in a separate arm of the U. The curr5nt and potential leads from Tz (in series with T3), insulated by thin glass tubes, pass lengthwise through the mica cross of Ta and then, with the potential leads from Ts, are threaded through small holes in mica strips which carry them out of the calorirneter, in the direction of the flowing gas, to hardsoldered connections with tungsten wires sealed through the glass walls. The temperature of the exit lines is adjusted so that the vapor passes through three-way stopcock 1 a t 20" above the boiling temperature of the compound. This stopcock can direct the vapor t o the flow calibration system or to the vaporizer. It is specially ground and is lubricated with a thin film of lithium stearate grease,13 which retains its consistency to a t least 100'. The flow calibration unit consists of a condenser, a detachable double trap E for collecting samples for timed intervals, a protective trap and stopcocks 4 and 5 . The lubricant usecl on the ground-glass joint above stopcock 2 is insoluble in hydrocarbon^.'^ The stem of stopcock 1 h;2s attached to i t a stainless steel sector which dips into a pool of mercury when the stopcock is turned, thus completing the circuit to an electric timer. The time interval, usually five or six minutes, is measured with an estimated accuracy of * 0.05 sec. The platinum resistance thermometers are of the coiled-filament strain-free type15and have icepoint resistances of about 24 ohms. The five thermometers were compared, a t the ice point and at other temperatures covering the range of the experiments, with a platinum resistance thermometer calibrated and certified by the National Bureau of Standards. During calibration the assembled calorimeter was mounted in the therinostated bath and a slow stream of helium was passed over thermometers TI, Tz antl TBto promote thermal homogeneity. From the calibration experiments constants of the Callendar equation were derived for each thermometer. 1-ma.measuring current is used for the five resistance thermometers connected in series. The power inputs to the vaporizer and calorimeter are determined by measuring the currents through the heaters and the potentials across them. A White double potentiometer, in conjunction with suitable calibrated resistors, standard cell and high-sensitivity galvanometer, is usetl for all electrical measurements. (13) I. E. Puddington, THISJOURN.4L, 65, 990 (1943). (14) T. P. Sager, I n d . Eng. Cheni., A n a l . E d . , 4, 388 (1932). (15) C. H. Meyers, Bureau Standards J . Research, 9, 807 (1932).

Jan., 19/47

HEATCAPACITIES

OF

WHEPTANE AND 2,2,3-TRIMETHYLBUTANE

Method of Measurement: (a) Rate of Flow and Heat of Vaporization.-The detachable double trap E, previously evacuated and weighed, is placed in position. Air is pumped from the connecting tubes and helium admitted t o the collecting system a t a pressure equal to that in the other parts of the apparatus. Then, with the vapor smoothdy cycling, stopcocks 2, 3 and 4 open, and suitable cooling agents on all traps, stopcock 1 is turned, starting the electric timer and diverting the hydrocarbon into the receiver for the desired interval. The vaporizer heater potentials and currents are recorded a t regular intervals, and the temperature of the issuing vapor is measured on thermocouple TC2 a t the beginning and end of the run. On completion of the collection, the hydrocarbon is frozen with liquid nitrogen and the helium pumped off. The receiver is detached and weighed, using a glass tare of equal volume and calibrated weights. No buoyancy corrections are necessary. The mass collected varied from 25 to 90 g. The total electrical energy delivered to the vaporizer during the collection period is obtained from the measured time interval and from interpolated mean values of the potential and current. The drift in the power input, from the storage batteries, is always very uniform and during a ten minute period is usually 0.01 to 0.057,. The uncertainty in interpolation is probably less than the absolute accuracy of the energy measurement, which is about 0.027,. Three corrections must be applied to obtain the total energy involved during the vaporization of the collected sample. (1) The boiling temperature, as measured on TC2, m:ty increase or decrease by as much as 0.05' between the beginning and end of the collection period, The change in the heat of vaporization with temperature may be neglected ; but because of the large masses involved, the change in temperature of the liquid and of the glass vessel causes the transfer of significant quantities of heat which can be calculated from the observed change in temperature, the known specific heats, the mass of glass and the average mass of liquid. Since the temperature change is small the expression (heat capacity glass heat capacity liquid) AT gives a sufficiently accurate estimate of the energy to be subtracted from the measured energy. The total correction ranges from 0 to 20 calories per mole. (2) m'hen the hydrocarbon is boiling, a temperature gradient exists between the surface of the liquid ,and the level of the heater. This gradient depends on the increase of pressure with depth and its value may be calculated from the density of the liquid and vapor pressure data or it may be measured by observations of temperature with TC:, a t different depths in the liquid. When m l grams of hydrocarbon are vaporized, the change of state involved is the conversion of nzl grams of liquid, at the mean temperature of the heater level, into ml grams of vapor a t the tem-

+

25

perature of the surface. This meaiis that the heat of vaporization, as determined a t the temperature of the surface, must have a correction applied for the transfer of ml grams of liquid from the heater level temperature to the surface temperature. In addition all parts of the wall of the containing vessel decrease in temperature by a uniform amount AT2 determined by the change in depth of the liquid above the heater. The form of the correction necessitated by these two depth factors is n z ~ CAT, ~, mzC,,ATz where ml, C,,, m2 and C,, refer to the mass and heat capacity of the hydrocarbon and glass involved, respectively, AT1 is the difference between the mean heater-level temperature and the surface temperature and AT2 is the change in wall temperature caused by the change in depth. The first term is the more important. The correction varies from 10 to 33 cal. per mole, depending on the fullness of the vaporizer, the amount removed and the pressure a t which the experiment is performed. (3) The current leads inside the vaporizer are insulated with loosely woven fiberglass sleeves which, due to capillarity, are wet by the hydrocarbon. The correction for unmeasured energy in the leads must therefore include both the immersed part and the wet portion of the leads above the surface. The sum of these two parts is constant a t a given temperature. The correction is about 25 cal. per mole. Having applied the three corrections described, the heat of vaporization is calculated from the relationship A1Zv = Q ( l - v / V ) , / M where Q is the corrected energy delivered to the vaporizer while ;21moles are being collected and the factor (1- (v/ 17)) makes allowance for the fact that the space vacated by the evaporating liquid becomes filled with vapor which is not condensed and weighed. The molal volume of the vapor V may be obtained with sufficient accuracy from the Berthelot equation and u, the molal volume of the liquid, is derivable from density data. To illustrate the combined effect of the corrections applied the data given for n-heptane a t 58.06' (see Table I) would have a precision of + 30 cal. per mole with no corrections applied instead of the * 2 cal. per mole reported. The precision indicated is a measure of the extreme range of the measured values. The apparent heat capacities determined during flow-calibration experiments check to within 0.0.55; of those determined under normal conditions when the hydrocarbon cycles and the liquid level in the vaporizer remain constant. This is good evidence for the validity of special corrections made during the flow measurement. Once proportionality between power input and mass rate of flow has been estabjished for a given boiling temperature, a determination of power input provides a measure of flow rate. Experiments show this direct proportionality to be independent of flow rate and of the duration of the flow measure-

+

GUY WADDINGTON, SAMUEL S. TODD AND HUGHM. HUFFMAN

2 (3

Vol. 60

ment if the boiling temperature is kept the same a t the different rates of flow. (b) Heat Capacity Measurements.-Apparent heat capacities are 'determined a t four or more flow rates from observations of the temperature increase produced by a measured power input to the calorimeter heater. Resistance thermometers TZand TBgive independent values of the apparent heat capacity a t each flow rate. A plot of these values against the reciprocal of the flow rate is linear, the mean deviation from linearity in this work being about f 0.05% (see Fig. 2). Extrapolation to zero value of the reciprocal corrects for the heat losses which even a t the highest temperature and lowest flow rate are less than 2%.

the measured power input are then obtained from the final .and initial temperatures observed on thermometers TBand TS(;.e., Ti final-T2 initial). A small correction is applied if the mean bath temperature drifts during the measurements. Only one thermometer enters significantly into the determination of an individual AT value, thus tending to eliminate errors due to differences in thermometer calibrations and to Joule-Thompson cooling caused by the pressure drop through the calorimeter. Under conditions of low temperature, low pressure and high flow rate, the latter effect would cause significant errors. Over the 8' temperature rise used in practice, the temperature coefficient of the Joule-Thompson cooling is such as to introduce an effect of measurable magnitude under some conditions but an adequate correction is applied. Experiments have shown that the apparent heat capacities are independent of the magnitude of the AT used, provided the mean temperature of a series of measurements is the same. From the previously determined relationship between power input to the vaporizer and mass of vapor flowing the flow rate is determined by a measurement of power supplied to the vaporizer. 57.0 Small corrections are applied for the energy generated in the immersed portion of the current 0" 46 5 leads and for the fact that the liquid returned to 357 hCZ the vaporizer carries some heat with it due to its being superheated by approximately 0.2'. The calorimeter power input is corrected for the 46.0 unmeasured energy generated in the leads a t the 0 5 10 15 20 entrance to the calorimeter. Experiments have 1 / F , min./mole. shown this effect to be about 0.03y0of the total. Fig. 2.-The apparent heat capacity of n-heptane as a During a given series of experiments the temfunction of flow rate. perature and hence the pressure in the vaporizer Actual measurements are made by observing are brought. to a chosen value by adjusting the temperatures on resistance thermometers TI, TB controlled helium pressure a t each flow rate. and TBwith vapor flowing but before heating the Over the range of flow rates and pressures studied, vapor and then repeating all observations after the pressure drop between the vaporizer and the the measured power input to the calorimeter has condensers varied from 1 to 33 mm., being a t a had enough time to bring the entire calorimeter to maximum for the highest flow rate and lowest a steady state. The time required to attain a operating pressure. From the geometry of the steady state is 3 to 15 minutes, depending on the system and a knowledge of the total pressure flow rates, which range from 0.05 to 0.25 mole per drop the actual pressure in the calorimeter can minute. The temperature of the bath is meas- be closely estimated. When this pressure differs ured with a resistance thermometer a t the be- significantly from the pressure in the vaporizer, a ginning and end of the experiment. The reading correction is applied to the measured heat capacity of each thermometer is recorded a t regular inter- to make its value correspond to the vaporizer vals during the experimental period. Thermom- pressure. This correction involves a measured eter readings on Tz and TB are made a t fifteen- value of ( AC*/ Ap)r and the pressure drop besecond intervals over a four-minute period and tween the vaporizer and the calorimeter. Its the results averaged. The range of the readings maximum value is O.1y0. Heat capacity values calculated from data obcaused by slight irregularities in flow rate, is +0.003 to *0.008 deg. Small thermal effects in tained with thermometers Tz and T3 agree on the the thermometer circuit, caused by thermal average to within O.O?i7, when operating condigradients in the heated vapor after it leaves the tions are closely controlled. Materials.-The n-heptane, obtained from a calorimeter, are measured with no current flowing through the thermometers and appropriate cor- commercial source, was distilled in a 77-plate rections are applied to the observed temperatures. column at about 3 cc. per hour at a reflux ratio The actual temperature increments produced by of about 4OO:l. From its melting point curve

1

1

1

-0 '-cPu"uc-iP

NO J

0's

-Hc9YaYI,rP

Jan., 1947

HEATCAPACITIES OF n-HEPTANO AND 2,2,3-TRIMETHYLBuTANE

the sample was found to have a purity of 99.97 mole per cent., assuming the impurity to be liquidsoluble, solid~-insoluble ; it had the constants d204 0.6837; n2"D 1.3877. The 2,2,3-trimethylbutane, obtained from General Motors Corp., was distilled similarly t o the n-heptane. A. melting-point analysis by the National Bureau of Standards showed its purity to be 99.69 mole per cent. Its density and refractive index were: d204 0.6900, n20D 1.3894. The most probable impurity, 2,4-dimethylpentane, since it i:j isomeric with triptane, would have a negligible effect on the measured heat capacities. Heats of Vaporization.-If corrections are made for all (energy exchanges, flow calibration experiments will yield accurate values of heats of vaporization, provided the boiling gives an essentially dry vapor. The absence of spray was suggested by the fact that the heat of vaporization was not affected by change of flow rate or by the presence or absence of a glass-wool spray trap. Pitzer,lo using the same method on other compounds, obtained reasonable agreement with literature values of heats of vaporization. The precision of the present work is higher than that reported by Pitzer, but literature values suitable for checking accuracy are not available at any of the temperatures studied. Table I gives the results of six separate measurements of the rate of flow and the corresponding heats of vaporization for 2,2,C:-triniethylbutane a t 80.80'. The other heats of vaporization given in the table are the average values of 3 to 6 determinations. The corrections outlined earlier in this paper were applied systematically. The precision is approximately * O.lyo. However, pending further studies, the accuracy is tentatively estimated as + 0.3%.'j" The Clausius-Clapeyron equation AHv = TAV (dpldt) may be used to calculate heats of vaporization if accurate values of vapor pressures and molal volumes of vapor and liquid are available. Experimental evidence obtained in heat capacity .measurements a t different pressures indicates that the gas imperfection predicted by the Berthelot equation is too small for the compounds studied in the temperature range of interest, hence it has not been used here to obtain the mola.1 volume of the vapor. In the case of nheptane parametersla of the Kellogg equation17 were available and have been used to calculate (l5a) Since the writing of this paper a measurement of the heat of vaporization of benzene has been made at 80.10'. The mean of three experiments is 7,348 * 2 cal./mole which is within 0.07% of t h e reliable value 7,353 cal./mole derived from the work of Fiock, Ginnings and Holton, J. Research Natl. Bureau a j Standavds, 6. 881 (1931). This result suggests that the accuracy of the heats of vaporization reported far n-heptane and 2,2.3-trimethylbutane may be in the neighburhood of *O.l%. (16) Unpublished values derived from t h e experimental work of Smith, Beattie and K a y (see ref. 22) were courteously supplied by the M ,W. Kellogg Co. (17) M. Elenedict. G. B. Webb and L. C. Rubin, J . Chcm. Phys.. 8 , 331 (1940).

24

TABLE I THEHEATSOF VAPORIZATION OF 2,2,3-TRIMETHYLBUTANE AND %-HEPTANE 1 cal. = 4.1833 int. ioules: at. wt. carbon = 12.01; at. w t . hydrogen = 1.008: P , mm.

758.4

195.7 597.7 391.2 195.2

Flow

AH" Clapeyron

AHv cal./mole

moledmin.

T,' C .

2,2,3-Trimethylbutane 80.80 0.056 6912 ,068 6921 .069 6914 .IO3 6918 .226 6918 .228 6924 Mean 6918 * 6 40.68 6 expts. 7461 * 4 90.48 77.33

58.06

%-Heptane 3 expts. 7715 * 2 3 expts. 7938 * 5 4 expts. 8244 * 2

7734 7945 8245

molal volumes of the vapor which when combined with molal volumes of the liquid and dp/dt derived from vapor pressure datals yield the heats of vaporization given in the last column of Table I which agree with experiment to better than 0.3%. The same method of calculation applied to propane, n-butane and n-pentane gives similar agreement with literature values of the heats of vaporization. If systematic errors in the heats of vaporization should be present, these may have no effect on the accuracy of the heat capacities since it is only necessary to establish a proportionality between mass of vxpor passing through the calorimeter per unit of time and electrical power supplied to the vaporizer. Experimental Heat Capacities.-Table I1 gives values of the heat capacities of n-heptane and 2,2,3-trimethylbutane a t the temperatures and pressures studied. In obtaining ( AC,/ Ap)T values from determinations a t one temperature and two or more pressures constant absolute errors in C, will cancel. For this reason results are reported to the nearest 0.005 cal./deg./mole since the precision of the results suggests that

TABLE I1 EXPERIMENTAL HEATCAPACITIES, CAL./DEC./MOLE 1 cal. = 4.1833 int. joules; at. wt. carbon = 12.01; at. wt. hydrogen = 1.008. T o , K.+357.10

373.16

n-Heptane 400.40 434.35

466.10

P , mm. 195.5 391.2 597.7

46.140 46.515

T o , K.-~X?8.80

P,mm. 195.7 758.4

43.085

50,555

53.965

57.0i5

54,200

57.215

2,2,3-Trimethylbutane 348.85 369.20 400.40

434.30

461.80

54.615 54.Ylj

57.515

47.805 48.095 48.410

45.345

50.930

47.585 48.1:jO

'

51.020

51.315

(18) A. P. I. Research Project at the National Bureau of Standards.

"Selected Values of Properties of Hydrocarbons." Table 2k

(Part 2), paraEins C?,dated May 31, 1944.

GUYWADDINGTON, SAMUEL S. TODD AND HUGHM. HUFFMAN

28

differences between measured heat capacities may possess higher accuracy than the quantities from which they are derived.

Vol. 69

temperature, and plotting C* vs. p, a linear extrapolation to zero pressure may be made, thus giving Ci and (AC,/ A+)T without recourse to any equation of state. This was done at all temperatures for n-heptane and at three temperatures for 2,2,3-trimethylbutane. The linearity of the three n-heptane points a t 373.15OK. is considerably better than the precision of the heat capacity measurements ( * O.lOjo) from which it is inferred that the use of only two pressures at the other temperatures will not lead to serious error. TABLE I11 HEATCAPACITIES IN THE IDEALGASSTATE CAL./DEG./MOLE 1 cal. = 4.1833 int. joules; at. wt. carbon = 1'2.I)l; a t . wt. hydrogen = 1.008 -C:, - C; cal.,/deg./mole/atm.T,O

441 340

! 360

380

400

420

440

460

T , OK. Fig. 3.-The heat capacity of n-heptane. The lower curve is for the ideal gas state and the upper curve for 1 atm. pressure The dashed portion is not experimentally realizable.

The upper curve of Fig. 3 shows values for nheptane of Cj,the heat capacity of the vapor a t 1 atmosphere pressure, as a function of temperature. Comparison with the work of others shows Pitzer'sIg one experimental point a t 423'K. to be about ?:l higher while Eucken and Sarstedt20 using a quite different variation of the flow method report a value a t 430°K. more than 37, higher than the results reported here. Bennewitz and Kossner's" value a t 410'1;;. is a b u t 10(;6 lower than that found in the present work. Correction for Gas Imperfection.-C;, the heat capacity of the vapor in the ideal state, is usually obtained from the experimental data by application of the relationship C, - Ci = - T v ( d ? i 'ZIT*)$ ~ rlp or other equivalent thermod o

dynamic equations. Frequently the second derivative has been evaluated from the Berthelot equation of state. This procedure may be aticquate for simple molecules, since for them the correction is si11al1, but ilil the present work with C7 molecules it becaine evident that either an accurate equation of state would be necessary for each compound studied or that another method of making the correction should be used if the Ci values were to retain the accuracy of the experimental data. By experimentally determining the heat capacities a t two or more pressures a t the same (19) Kenneth S. Pitzer, THISJOURNAL, 62, 1224 (1940). ( 2 0 ) A . liucken and R. Sarstedt. Z. p k r s i k . Cizem., B50, 143 (1941). (21) K Hrnnewit? a n d W . Rossner. i h i i f . . B39. 126 119%)

K.

C;

C:

Expt.

Berthelot

357.10 373.15 400.40 434.35 466.10

47.22 48.66 51.08 54.30 67.27

%-Heptane 45.77 1 . 4 5 0.65 47.51 1 . 1 5 .57 .71 .46 50.37 ,45 .36 53.85 57.00 .27 .29

328.80 348.85 369.20 400.40 434.30 461.80

44.08 46.07 48.13 51.32 54.82 57.52

2,2,3-Trimethylbutane 42.71 (1.34) 0.71 .60 45.09 ( .98) 47.39 .74 .50 50.92 .40 .40 .31 54.54 .28 57.36 ( .16) .26

BeattieBridgman

L;ellogg

0.32 .29 .25 .20

1.28 1.06 .79

.I7

.42

,57

Table 111 gives values of Cp', Ci and C j and

Cy - Cp" for n-heptane and 2,2,,3-trimethylbutane. Cj and Cp' were obtained by use of the experimental values of (Ac,lAp)~. Since the values of Cj - C j obtained experimentally are considerably larger than those predicted by the Berthelot equation of state, it becomes of interest to examine results obtained from equations of state based on experimental data. This is possible for n-heptane since parameters of the Beattie-Bridgman22 and Kellogg16equations have been evaluated for this compound. The Berthelot equation results were obtained from the relationship Cj - Ci = 81RT:P/32PcT3 using 540.17'K. and 27.00 atm. for the critical temperature and pressure of nheptane2?while for 2,2,3-trimethylbutane531.5'K. and 29.73 atm.24were user$. The Beattie-Bridgman correction utilized the approximate relationshipz5

c; - CO = 2'4 -2 + 12c RTz

T4

The K e l l ~ g g l ~ equation ~~' was used in the suffi(22) L. B. Smith, J. A. Beattie and W. C. Kay, THISJOURNAL, 69, 1587 (1937). (23) J. A. Beattie and W. C. K a y , ibid., 59, 1586 (1936). (24) Graham Edgar and George Calingaert, ibid , 51, 1.540 (1929). (2.5) J. A Reattie. Phus RPV..34. 1615 (1927)

HEATCAPACITIES OF %-HEPTANE AND 2,2,&TRIMETHYLBUTANE

Jan., 1947

ciently exact form P = RTd

+

+

+ (BoRT - Ao -

Co/T2)d2 (b.RT - a c / T 2 ) d3. From this equation Cj - C i is readily obtained by use of the thermodynamic relationships

29

involved the determination of three heats of vaporization ranging from 7,715 to 8,244 cal.1 mole. This precision has been confirmed by other work now in progress. C: for the two conipounds studied may be represented by the empirical equations

+

CE

n-heptane: Ci = -0.2365 0.14864T-0.00005583 T2 2,2,3-trimethylC j = 1.800 0.153201T-0.000053907T2 butane:

+

- Cl = R

This procedure avoids the cumbersome second All experimentally derived points are fitted by derivative of volume with respect to temperature these two curves to f0.05%. The triptane point a t constant pressure. a t 461.SO’K. obtained by the extrapolation The differences between the results from the method previously described deviates from the three equations suggest that caution should be equation by 0.16’%. The precision of the measused in obtaining C: from experimental data by urements is such that a clear-cut distinction is the use of equations of state, particularly a t low made between the heat capacities of the two isotemperatures. For n-heptane a t 3T3.15’ the er- meric compounds. While the precision of the measurements is ror in C i when the Berthelot equation is used is satisfactory the question of accuracy must be about 1.3%. The increase of Ci - Cp” with decrease in temperature is predicted reasonably well approached with caution. I n the dynamic sysby the Kellogg equation but the agreement is not tem employed, it is difficult to be sure that proper corrections have been applied for all variables perfect. which influence the final values. In principle Accurate .experimental values of ( A C , / A ~ ) T provide a useful check on the assignment of pa- the accuracy of the calorimetric measurements rameter values in the construction of empirical can be checked by comparison with statistically equations of state. This is particularly true in calculrtted values. However, the increasing unthe low-temperature, low-pressure region where certainties and incompleteness of spectral analysis PVT data are not ordinarily available. In addi- and the greater gas imperfections as molecular tion a knowledge of ( A C p / A p ) r as a function of complexity increases make accuracy a matter to temperature gives information concerning the be resolved through the accumulation of data, by temperature dependence of the second virial co- varying experimental methods in different laboratories, which will possess internal consistency and efficient since (AC,/AP)T = - T(b2B/bT2jp be consistent with other related thermodynamic where B is defined by p V = RT Bp. For 2,2,3-trirnethylbutane heat capacities were quantities which have been determined indemeasured a t one pressure only a t 337.10, 373.15 pendently. Acknowledgment.-We thank Richard 34. and 4GG.10°K. The - Cp” values (it1 parenGooding of this Laboratory for the distillation theses in Table 111) a t these points were obtained by an extrapolation method from the equation and measurement of the physical constants of (AC,/Ap: T = -- T(b2B/bT2)p with B being of the the two compounds as well as for the estimate of form Bo .- ( A , / T )- ( C / T 3 ) . This type of equa- impurity in the n-heptane by the melting point tion fits the data from both compounds and gives method. We are grateful to F. D. Rossini of the National Bureau of Standards for determinti linear plot of T 2( A C p / A P ) r as. l / T 2 . ing the purity of the 2,2,3-trimethylbutane and The extrapolated values of Cp’ - Cp” obtained t o General Motors Corp., which supplied the from this plot a t the temperatures mentioned sample of this compound. appear to be only slightly less reliable than measured values a t other temperatures. Summary

+

Discussion The emphasis in this investigation has been on accuracy. Llnanalysis of the many known variables entering into the measurements indicates that the precision of the results should be in the neighborhood of +O.lyo and this is checked not only by the reproducibility of a given point but by internal consistency of a high order. For example the heat capacities of n-heptane a t 373.15’K. and pressures of 193.5, 391.2 and 597.7 mm., when plotted against pressure, deviate from the linearity expected for a low pressure vapor by less than 0 0 3 ; w e n though the measurements

An improved constant-flow calorimeter for vapors IS described. Details involved in the experimental procedure and in the treatment of the data are outlined. Results obtained have a precision of =tO.lyo. The heat capacities of Iz-heptane and 2,2,3trimethylbutane have been measured a t several pressures over the temperature range 330 to 465OK. From measured values a t two or more pressures a t the same temperature the heat capacity of the vapor in the ideal state is obtained by linear extrapolation to zero pressure. It is believed that this method of making the cw-rec-

PAULJ. FLORY

30

tion for gas imperfection is more accurate than the use of available equations of state. C: for the two compounds studied may be represented by the following empirical equations with a mean deviation of less than 0.1% n-Heptane C i = -0.236 2,2,3-Trimethylbutane: C i = 1.800

+ 0.14864T - 0.000055583P + 0.15320T - 0.000053907T2

[CONTRIBUTION

Vol. 69

The method of measurement also yields precise heats of vaporization. For n-heptane the heats of vaporization reported a t 90.48, 77.33 and ri8.00’ are 7,715, 7,938 and 8,244 cal./mole respectively while for 2,2,3-trimethylbutane the values a t 80.50 and 40.68’ are 6,918 and 7,461 cal./mole. BARTLESVILLE, OKLAHOMA RECEIVED AUGUST8, 1946

NO. 134 FROM THE GOODYEAR RESEARCH LABORATORY]

Molecular Size Distribution in Three Dimensional Polymers. V. Post-gelation Relationships BY PAULJ. FLORY In previous papers1-* of this series statistical methods were applied to the structural interpretation of three dimensional polymers, Le., polymers some of the structural units of which are polyfunctional in the sense that they connect with more than two other neighboring units of the polymer, Critical conditions for the occurrence of gelation were explored, and methods were set forth for computing the extent of reaction at the gel points in condensation processes involving polyfunctional units1 or in addition polymerizations accompanied (or followed) by crosslinking reaction^.^.^ The theory of gelation resulting from cross-linking was extended by Stockmayer6 and by Walling.6 Gelation occurs in three dimensional condensation polymerizations a t somewhat higher extents of reaction than predicted by theory owing to intramolecular condensation^^^^ which are not taken into account by the theory.* The discrepancy generally is small, and the essential correctness of the statistical theory is adequately confirmed by the correlation between observed and calculated gel points. The writer applied the statistically derived molecular size distribution relationships beyond the gel point as well as t o polymers which had not been carried to gelation. By appropriately summing over the distribution functions for all finite species, i t was possible to derive the weight fraction of sol, average molecular weights of the sol, and the concentration of cross linkagcs in the (1) P. J. Flory, THIS JOURNAL, 63, 3083 (1941). (2) P. J. Flory. ibid., 63, 3091 (1941). (3) P. J. Flory, ibid., 63, 3096 (1941). ( 4 ) P. J. Flory, J . P h y s . Chem., 4 6 , 132 (1942). (5) W. H. Stockmayer, J . Chem. Phys., 12, 125 (1944). (6) C. Walling, THISJOURNAL, 67, 441 (1945). (7) W. H. Stockmayer and L. L. Weil, unpublished. See “Advancing Fronts in Chemistry,” Reinhold Publishing Corp., S e w York, N. Y . , Chapter 6 by W. H. Stockmayer. (8)A theory which will take into account intramolecular connections obviously is needed but, as Stockmayer has pointed out (paper given before the Division of High Polymer Physics of the American Physical Society, January, 1946), the mathematical difficulties a1)pear to be insuperable.

sol; gel properties were obtained b ~ difference. r Stockmayerg generalized and extended the size distribution relationships, but was hesitant to apply them beyond the gel point. His criticisms of the writer’s efforts in this direction centered upon the approximation2s3 according to which intramolecular (“cycIic”) connections in finite (sol) molecules are neglected while acknowledging that they must be present in the infinite network (gel) structures. On the other hand, the derivation of the size distribution equations applies equally to the sol existing beyond the gel point, or to the polymer preceding gelation.I0 The distribution equations are oblivious of the gel (except by omission), hence the frequrnt occurrence of intramolecular connections in thc gel does not constitute a violation of the basic approximation. n‘hile the logic of this procedure appears to be satisfactory, it must be admitted that derivation of gel properties by difference is not as straightforward as might be desired. In this paper polymers randomly cross-linked to the extent that they contain a gel fraction will be treated by a more general procedure applicable to any distribution of chain lengths. The method here employed is independent of size of complexity distribution ryuations (except in calculating weight average molecular weight) ; i t is more direct and simpler than the proccdures previously applird to the problem of post-gelation constitution. The results obtained confirm and extend previous conclusions. Theoretical Treatment Molecular size distribution function^^.^^^ for three dimensional polymers may show a single maximum a t relatively low molecular weight, but beyond this maximum they decrease monotonically toward zero. This is true after as well as before gelation. It is evident, therefore that an intermediate fraction between finite sol molecules ( 9 ) W H. Stockmayer, J. Cham. P h y s , 11, 45 (1943) ( 1 0 ) P J Flory, Chem. Rev., 39, 137 (19461.